Coronary Artery Disease

Women develop heart disease about 10 years later than men do, but in women this disease is, as it is for men, the number-one cause of death. Smoking is clearly a major risk factor for heart disease in both sexes, and most physicians already devote substantial attention to smoking cessation counseling. In 1997, studies focused on three other risk factors for heart disease in women: diabetes, hypertension, and hypercholesterolemia.

Women with Diabetes Had Lower High-Density Lipoprotein Cholesterol Levels

U.K. Prospective Diabetes Study 27: plasma lipids and lipoproteins at diagnosis of NIDDM by age and sex. Diabetes Care. 1997; 20:1683-7.

Diabetes is a well-known risk factor for the development of coronary artery disease. However, women with diabetes who have a myocardial infarction are more likely to die than are diabetic men who have the same event [1]. It has been suggested that this higher mortality rate may be partially due to differences between the sexes in lipid profiles.

In this cross-sectional study, the lipid profiles of 2139 men and 1574 women with newly diagnosed type 2 diabetes were compared across age groups and between the sexes; patients ranged in age from 25 to 65 years and were otherwise in good health. These patients were also compared with a group of British persons who did not have diabetes and with participants in the U.S. Lipid Research Clinics Population Prevalence Study.

Only 5% of the women were taking estrogen replacement or oral contraceptives. The women were more obese than the men and had higher fasting insulin levels, higher hemoglobin A_1c levels, worse glucose control, and greater systolic and diastolic blood pressures. Lipid profiles are shown in Table 1. Women with diabetes had the highest levels of both total and low-density lipoprotein (LDL) cholesterol. Moreover, compared with the levels of age- and sex-matched persons without diabetes, high-density lipoprotein (HDL) cholesterol levels were 23% lower in diabetic women but only 9% lower in diabetic men.

When age was factored in, it was found that lipid levels in women progressively worsen until age 50 years and then plateau until at least age 65 years. Multivariate analyses found no associations between lipid levels and obesity, high blood pressure, glucose levels, or body mass index.

The authors suggest that the lower HDL cholesterol levels seen in these diabetic women may explain as much as 30% of their increased risk for coronary artery disease. Several important treatment studies in diabetic women are not yet complete, but subgroup analysis of the Scandinavian Simvastatin Survival Study [2] suggests that aggressive management is beneficial. Counseling to improve metabolic control through weight loss and aerobic exercise may also alleviate lipid abnormalities.

Hypertension Treatment Prevented More Myocardial Infarctions Than Strokes

Gueyffier F, Boutitie F, Boissel JP, et al. Effect of antihypertensive drug treatment on cardiovascular outcomes in women and men. A meta-analysis of individual patient data from randomized, controlled trials. The INDANA Investigators. Ann Intern Med. 1997; 126:761-7.

Coronary artery disease is the most common, most deadly sequela of hypertension, yet patients often think of stroke prevention as the reason to treat hypertension. This study compared differences between the sexes in the effects of antihypertensive treatment on cardiovascular outcomes. Included in this reanalysis of individual patient-level data were 7473 older (>59 years of age) and 13 362 younger (30 or 35 years of age to 59 years of age) women as well as 5010 older and 14 933 younger men who had been entered into seven trials that primarily used thiazide diuretics and ß-blockers to treat hypertension. The outcomes measured included mortality, all strokes, all coronary artery events, and all cardiovascular events.

Overall, the rate of cardiovascular events was higher in the older groups than in the younger groups (as expected) and was greater for men than for women. Graphical representation suggested that differences between the sexes were related to differences in untreated risk (Figure 1). The results for women and men were most similar in studies that focused on treatment of systolic hypertension with thiazides and ß-blockers in the elderly. Although the treatment effect is greater for stroke than for myocardial infarction, stroke is much less common than myocardial infarction. Prevention of myocardial infarction is a major benefit of the effective treatment of hypertension in women as well as in men.

View larger version (25K): [in this window] [in a new window]   Figure 1. Sex-specific major cardiovascular events in clinical hypertension trials. Older persons were older than 59 years of age; younger persons were 30 or 35 years of age to 59 years of age. White bars represent untreated women; bars with stripes slanting up from left to right represent treated women; bars with stripes slanting down from left to right represent untreated men; black bars represent treated men. C&W = Coope and Warrender; HDFP = Hypertension Detection and Follow-up Program; MRC = Medical Research Council; SHEP = Systolic Hypertension in the Elderly Program; STOP = Swedish Trial in Old Patients. Adapted from Gueyffier F, Boutitie F, Boissel JP, et al. Effect of antihypertensive drug treatment on cardiovascular outcomes in women and men. A meta-analysis of individual patient data from randomized, controlled trials. The INDANA Investigators. Ann Intern Med. 1997;126:761-7.

Type, Not Amount, of Fat Intake Predicted Coronary Heart Disease

Hu FB, Stampfer MJ, Manson JE, et al. Dietary fat intake and the risk of coronary heart disease in women. N Engl J Med. 1997; 337:1491-9.

Assessing the implications of fat intake can be confusing to both patients and clinicians. Overall, most experts recommend eating less fat, especially less saturated fat. Butter and meat, which have the greatest concentrations of saturated fats, are the most deleterious common fats; these are followed by hard margarine, which is predominately transunsaturated fat. Peanut butter is a mixture of monounsaturated and saturated fats; olive and canola oils are primarily monounsaturated fats. The most protective fats are corn and sunflower oils, which are largely polyunsaturated fat.

Despite these recommendations, we have not known the risks of intake of specific fats with respect to the health of women's coronary arteries. The Nurses' Health Study, a cohort study that began in 1980, sought to end that uncertainty. More than 80 000 nurses completed an initial, detailed food questionnaire and were resurveyed in 1984, 1986, and 1990. They also completed general medical histories every 2 years. Intake of specific fats was divided into quintiles according to the degree of saturation of fats. The cohort was followed for 14 years; the clinical outcomes assessed included non-fatal myocardial infarction and death from coronary artery disease.

The results sustained some intuitive beliefs. First, intakes of specific fats were interrelated. For example, persons who ate more olive oil ate less margarine. Second, the greater the intake of nonfat foods, the less the intake of fat overall.

However, after 14 years of follow-up and after adjustment for confounding variables in the multivariate analysis, total fat intake did not correlate with clinical outcomes but specific fat intake did. The worst outcomes were found in association with higher intake of transunsaturated fats (hard margarine); a 2% increase in intake almost doubled the clinical risk. A 5% increase in saturated fats produced a 15% increase in adverse outcomes. Vegetable fat fared better; a 5% increase in monounsaturated fat intake was associated with a 10% decrease in adverse events, and a 5% increase in intake of polyunsaturated fats produced a 38% decrease in adverse events.

In this cohort, after careful multivariate analysis, total fat intake did not seem to predict coronary artery events. However, vegetable oils, especially olive and corn oils, seem to reduce risk for these events, whereas hard margarine, meat, and butter are especially harmful. Because this study is not a prospective, interventional clinical trial, we do not yet know whether changing the type of dietary fat ingested will change the risk for adverse outcomes. The Women's Health Initiative should answer that question when its clinical trial, now in progress, is completed.

Hormone Replacement Therapy plus Pravastatin Effectively Improved Lipid Levels

Davidson MH, Testolin LM, Maki KC, et al. A comparison of estrogen replacement, pravastatin, and combined treatment for the management of hypercholesterolemia in postmenopausal women. Arch Intern Med. 1997; 157:1186-92.

Treatment of hypercholesterolemia, especially as secondary prevention, is effective in reducing the risk for coronary artery events in both women and men. However, only about 47% of women with known coronary artery disease receive treatment for their hypercholesterolemia [3, 4]. In contrast, much attention has been paid to hormone replacement therapy for the prevention of coronary artery disease.

Some observational studies, but few interventional trials, have compared the treatment of hypercholesterolemia, primarily with "statin" drugs, with hormone replacement therapy [5]. This clinical trial compared the effects of estrogen and pravastatin on the lipid profiles of postmenopausal women. Investigators randomly assigned 76 postmenopausal women with hypercholesterolemia to receive one of four treatment regimens: conjugated estrogen (0.625 mg/d), pravastatin (20 mg/d), both conjugated estrogen and pravastatin, or placebo. Changes in serum lipid profiles and compliance were assessed at 16 weeks.

The results are summarized in Table 2. After 16 weeks, the group receiving combined therapy had had the greatest positive effects on LDL cholesterol levels. Regimens that included estrogen, 0.625 mg/d, had a positive effect on HDL cholesterol levels. Levels of triglycerides were increased by estrogen and decreased by pravastatin.

From this relatively brief study, which used estrogen without progesterone, several conclusions can be drawn. First, estrogen improves both HDL and LDL cholesterol levels. Second, the improvement seems to be dose-related. Adding pravastatin further improves lipid levels. Data from the Postmenopausal Estrogen/Progestin Interventions (PEPI) trial showed that adding progesterone to estrogen (necessary for women with a uterus) is more effective than placebo for hypercholesterolemia [6]. Trials are now in progress to determine whether clinical outcomes will predictably follow these positive changes in serum lipid levels.

Hormone Replacement Therapy

Hormone replacement continued to be a major clinical topic in 1997. For practitioners, questions focus on the risks and benefits of hormone replacement therapy with respect to heart disease, osteoporosis, and breast cancer. Should every postmenopausal woman receive this therapy? If not, who should and who should not receive it? Must a woman take the standard dosage of conjugated estrogen (0.625 mg/d), or can she take less? Are the estrogen alternatives as safe and beneficial as traditional regimens? The last two articles in this section examine issues related to potential adverse effects of hormone replacement therapy.

Hormone Replacement Therapy Was Most Effective in Women with Coronary Risks

Grodstein F, Stampfer MJ, Colditz GA, et al. Postmenopausal hormone therapy and mortality. N Engl J Med. 1997; 336:1769-75.

Most physicians believe that the benefits of hormone replacement therapy with respect to the bone and the heart outweigh the risks for cancer of the breast and endometrium, but large, long-term studies have been lacking. This study, part of the Nurses' Health Study, followed 121 700 nurses (who were 30 to 55 years of age in 1976) for 18 years.

The investigators documented 3637 deaths in the course of the study. Each woman who died was matched with 10 controls who were alive at the time of her death. Each biennial questionnaire asked about use of hormone replacement therapy. Current users of this therapy were compared with nonusers. To eliminate potential confounding factors, the authors adjusted for risk factors for heart disease, body mass index, exercise, vitamin E intake, fat intake, and known risk factors for breast cancer.

All-cause mortality was significantly lower in current users of hormone replacement therapy (relative risk reduction [RRR], 37% [95% CI, 30% to 44%]). However, among women who used this therapy for more than 10 years, this reduction in death was attenuated (RRR, 20% [CI, 4% to 33%]), primarily because of an increase in the incidence of breast cancer in long-term users. The greatest reduction in risk was in risk for death from coronary heart disease, and the women who benefited most were those without at least one risk factor for coronary heart disease (RRR, 49% [CI, 43% to 55%]). Women with no coronary risk factors benefited somewhat less (RRR, 19% [CI, 38% to –28%]).Past use of hormone replacement therapy was not protective: Once a woman had not taken the therapy for 5 years, her risk returned to that of a nonuser.

An important question is whether adding progestin to estrogen negates some of estrogen's cardioprotective benefits. In this study, the protection against coronary heart disease conferred by estrogen plus progestin was similar to that provided by estrogen alone.

Short-term users of hormone replacement therapy had no increased risk for breast cancer compared with nonusers. Even among women with a family history of breast cancer, no increase in breast cancer mortality was seen among those who had used hormone replacement therapy for 5 years or less.

At least five clinical conclusions can be drawn from this study. First, current use of hormone replacement therapy seems to be associated with lower mortality, particularly from coronary heart disease. Second, protection is greatest among women with the highest risk for coronary artery disease. Third, overall protective effects seem to be attenuated but are still present after 10 years, a finding that may make decision making difficult for patients and clinicians. Fourth, however, cessation of hormone replacement therapy returns a woman to the risk of a woman who never used hormone replacement therapy after only 5 years. Finally, the investigators tried to address the healthy-user bias. Past studies have been criticized because users of hormone replacement therapy generally tend to have healthier lifestyles; they may be better educated, more fit, and less overweight and may have fewer coronary risk factors. Because the Nurses' Health Study cohort was a relatively homogenous group and all of the women studied probably had relatively healthy lifestyles, healthy-user bias may have been minimized. By the same token, however, the generalizability of these results may be limited.

Other studies have found reduced mortality associated with use of hormone replacement therapy, even among elderly women [7]. However, a recent study from Kaiser Permanente showed no differences in the rate of first myocardial infarction among current users and nonusers [8]. The fact that this was a case–control study of women with myocardial infarction raises the possibility of recall bias. In addition, the study measured first coronary heart disease events, not deaths. Finally, the 95% CIs for risk reduction ranged from 0.66 to 1.40, a wide interval that is still compatible with meaningful protection by estrogen. The controversy will probably be settled when the Women's Health Initiative study is completed in 2006.

Decision Analysis Predicted Who Should Receive Hormone Replacement Therapy

Col NF, Eckman MH, Karas RH, et al. Patient-specific decisions about hormone replacement therapy in postmenopausal women. JAMA. 1997; 277:1140-7.

In the meantime, how can clinicians help a particular woman decide whether to take hormone replacement therapy? Population data are convincing: Breast cancer kills 43 000 women annually, but coronary artery disease claims 233 000 women and hip fractures kill another 65 000. A woman facing the decision may be at higher or lower risk for particular diseases or may have specific fears about developing cancer. Moreover, no data from randomized, controlled trials are currently available to aid in decision making.

Therefore, using a mathematical model, Col and colleagues undertook a decision analysis to examine the effect of hormone replacement therapy on life expectancy in postmenopausal women with different risk profiles for heart disease, breast cancer, and hip fracture. They used published regression models to estimate lifetime risks for coronary heart disease, breast cancer, hip fracture, and endometrial cancer. The effect of hormone replacement therapy on life expectancy was estimated for women with various risks for coronary heart disease, osteoporosis, and breast cancer.

According to this model, hormone replacement therapy should increase life expectancy for almost all postmenopausal women. The greatest gain, 41 months of extra life, was seen in the women who had the highest risk for coronary heart disease but the lowest risk for breast cancer. The only women who did not derive improved life expectancy from hormone replacement therapy were those who were at low risk for coronary heart disease and high risk for breast cancer.

In decision analyses, sensitivity analysis is done to examine the effect of uncertainty or disagreement about the estimates used in the model. In performing this sensitivity analysis, the researchers took into account the fact that hormone replacement therapy users who develop breast cancer tend to have less aggressive cancer. This model suggests that all women would live longer if they took hormone replacement therapy. Although this study used hypothetical patients, it may serve as a guide to advising actual women by sustaining the belief that hormone replacement therapy extends the lives of postmenopausal women who do not already have breast cancer. However, living longer may not be the only goal for women who use hormone replacement therapy. Other factors, such as quality of life, must also be considered.

Lower-Dose Estrogen Seemed Safer and Effective

Genant HK, Lucas J, Weiss S, et al. Low-dose esterified estrogen therapy: effects on bone, plasma estradiol concentrations, endometrium, and lipid levels. Estratab/Osteoporosis Study Group. Arch Intern Med. 1997; 157:2609-15.

The usual dosage of esterified estrogen prescribed for women receiving hormone replacement therapy is 0.625 mg/d. There has been concern that lower doses might not adequately protect against bone loss and that high doses might increase the risk for cancer, especially endometrial cancer.

A clinical trial of 406 postmenopausal women determined the effects of unopposed esterified estrogens on bone mineral density, lipid levels, and endometrial thickness. An additional goal was to correlate these effects with plasma estradiol levels. After receiving no estrogens for at least 8 weeks, all participants were given calcium supplementation and were randomly assigned to receive one of three dosages of esterified estrogens (0.3 mg/d, 0.625 mg/d, or 1.25 mg/d) or placebo for 24 months. To minimize potential confounding effects on bone density, no women received progestin. Bone densitometry, endometrial biopsies, and lipid analyses were done every 6 months.

Table 3 summarizes results for the 54% of women who completed the study. Women receiving higher doses of estrogen tended to discontinue treatment because of endometrial hyperplasia. Substantial numbers of those receiving placebo and the lowest estrogen dosage tended to drop out because of inadequate relief of menopausal symptoms or vaginal dryness. In the intention-to-treat analysis, women assigned to placebo lost 2.5% of their bone density, but those assigned to any estrogen dosage gained bone density. Low-density lipoprotein cholesterol levels were decreased and HDL cholesterol levels were increased in a dosage-related manner. Finally, the incidence of endometrial hyperplasia was increased in women receiving 0.625 mg or 1.25 mg of estrogen per day but not in women receiving 0.3 mg of estrogen per day.

In conclusion, low-dosage estrogen seems to be safe for the endometrium and may preserve bone to some degree, but it may not adequately alleviate the symptoms of menopause. Although low-dosage estrogen was associated with an increase in bone mineral density, it is not known whether the increase will be maintained and will translate into decreased risk for fracture with long-term use. Similarly, the effects of low-dosage estrogen on clinical coronary heart disease are unknown. For women who are unwilling or unable to take higher estrogen dosages and those who cannot tolerate or do not wish to take progestin, low-dosage unopposed estrogen may be an appropriate alternative.

Raloxifene Was Safe in the Short Term

Delmas PD, Bjarnason NH, Mitlak BH, et al. Effects of raloxifene on bone mineral density, serum cholesterol concentrations, and uterine endometrium in postmenopausal women. N Engl J Med. 1997; 337:1641-7.

Raloxifene is a selective estrogen-receptor modulator (SERM), one of a class of agents being developed in the hope of realizing the benefits of estrogen without the risks (endometrial stimulation and increased risk for breast cancer). Raloxifene is the first of several drugs of this class, and this trial of raloxifene therapy investigated several outcomes.

Investigators randomly assigned 601 healthy, postmenopausal European women to receive one of three dosages of raloxifene (30, 60, or 150 mg/d) or placebo. All women were 45 to 60 years of age. Ultrasonography was used to measure bone mineral density and endometrial thickness initially and at 6-month intervals; lipids were analyzed every 3 months. Follow-up continued for 2 years.

Only about 75% of the participants completed the study. The high drop-out rate occurred partly because raloxifene does not improve vasomotor symptoms. In the intention-to-treat analysis, bone mineral density was higher among women assigned to any raloxifene dosage than among women assigned to placebo. Total and LDL cholesterol levels both decreased in a dosage-related manner among those receiving raloxifene, but HDL cholesterol and triglyceride levels did not change. No major differences in endometrial thickness were found.

Several conclusions can be drawn from this study. First, raloxifene was associated with an increase in bone density, but the increase was not as great as that seen with estrogen. Second, although raloxifene did improve levels of LDL cholesterol, it did not increase HDL cholesterol levels, as estrogen does. Because HDL cholesterol is particularly important as a protective factor against coronary heart disease in women, raloxifene may not be as beneficial as estrogen for prevention of coronary heart disease. Because this study was relatively brief, clinical outcomes (such as incidence of coronary heart disease, fractures, and cancer) were not measured. In theory, raloxifene should not increase risk for breast cancer, but the risk with long-term use is unknown. In the meantime, raloxifene and other SERMs are exciting drugs that require long-term studies with clinical end points. Until such studies are completed, the role of these drugs in postmenopausal women should remain minor. Raloxifene might be considered for women who have firmly decided that they do not want to take hormone replacement therapy because of fear of breast cancer and are aware of the uncertain long-term benefits and risks of raloxifene treatment.

Transvaginal Ultrasonography Is Helpful When Results Are Normal

Langer RD, Pierce JJ, O'Hanlan KA, et al. Transvaginal ultrasonography compared with endometrial biopsy for the detection of endometrial disease. N Engl J Med. 1997; 337:1792-8.

Postmenopausal women who have vaginal bleeding traditionally undergo endometrial biopsy, a procedure that can be done in the office. Biopsies can be limited because they are done blindly (that is, they may not include critical endometrial samples). Moreover, even in very experienced hands, the procedure sometimes cannot be completed because of cervical stenosis and may cause cramps similar to menstrual cramps. For these reasons, transvaginal ultrasonography has been proposed as a way to detect endometrial hyperplasia, the precursor to endometrial cancer.

The 3-year-long PEPI trial was undertaken to test the cardiovascular effects of hormone replacement therapy on women who had recently become menopausal. As part of the trial, women had both endometrial biopsies and transvaginal ultrasonography. The investigators compared transvaginal ultrasonography with endometrial biopsy for the identification of endometrial hyperplasia in asymptomatic women.

A total of 596 women had a uterus, and 577 examinations were done on 448 of these women. Forty ultrasonograms were deemed uninterpretable, mostly because of uterine myomas. An endometrium at least 5 mm thick on ultrasonography was used as evidence of abnormality; the test characteristics are shown in Table 4.

Transvaginal ultrasonography was shown to be fairly accurate when its results were negative, at least in healthy women. However, it produces many false-positive results and thus is a poor screening test. For now, endometrial biopsy remains the gold standard for the diagnosis of endometrial problems in postmenopausal women, although ultrasonography may have a role in some symptomatic women. The most important gynecologic information from this cohort is that no woman with a uterus should be taking only estrogen, 0.625 mg/d or more, because more than one third of such women will have abnormal results on endometrial biopsy after 3 years.

Hormone Replacement Therapy-Related Risk for Breast Cancer Was Borne Out

Breast cancer and hormone replacement therapy: collaborative re-analysis of data from 51 epidemiological studies of 52,705 women with breast cancer and 108,411 women without breast cancer. Collaborative Group on Hormonal Factors in Breast Cancer. Lancet. 1997; 350:1047-59.

The risk for breast cancer associated with hormone replacement therapy is a major reason why postmenopausal women refrain from initiating therapy. Previous estimates of relative risk for breast cancer from various studies were as much as 30% greater for hormone replacement therapy users compared with nonusers. The goal of this comprehensive review was to more accurately define this risk.

Clinical and demographic data on 161 116 women from 51 studies of breast cancer done in 21 countries were assembled and analyzed for consistency. These data represent about 90% of the world's epidemiologic information on breast cancer risk associated with hormone replacement. The focus of the analysis was 53 865 postmenopausal women, 33% of whom had used hormone replacement therapy at some time.

The relative risk for breast cancer was 1.023 (CI, 1.011 to 1.036) for each year of hormone replacement therapy use (a risk similar to that of undergoing menopause 1 year later). For women who had used hormone replacement therapy for at least 5 years (mean, 11 years), the relative risk for breast cancer was 1.35 (CI, 1.21 to 1.49). However, 5 years after cessation of hormone replacement therapy, risk returned to baseline. The cumulative incidence of breast cancer for women 50 to 70 years of age who do not receive hormone replacement therapy is 45 per 1000. The cumulative excesses of breast cancer among 1000 women who used hormone replacement therapy for 5, 10, and 15 years were 2, 6, and 12 cases, respectively. Women with lower body mass indexes had a greater risk for breast cancer associated with hormone replacement therapy. Breast cancer associated with hormone replacement therapy tended to be less advanced than that detected in nonusers, perhaps because of more intensive follow-up surveillance.

These results confirm that the risk for breast cancer is increased in women who use hormone replacement therapy over the long term. However, overall mortality may still be lower for hormone replacement therapy users because coronary artery disease is so much more common than breast cancer. In fact, in a population-based cohort study of women with a family history of breast cancer, users had about 33% more breast cancer, but their overall mortality rate was lower than that of nonusers (RRR, 33% [CI, 11% to 49%]) [9]. The secondary finding of a greater risk among slender women requires confirmation because it was not reported previously. Finally, this study does not show whether the risk for breast cancer with estrogen alone is the same as that in women taking estrogen plus progesterone, because that information was not consistently available.

Although hormone replacement therapy is a well-established regimen for many women, many questions about the appropriate duration of this therapy remain unanswered. There is much to consider. Beginning this therapy early in menopause helps to prevent the rapid bone loss that starts at about that time. Yet the average age at which a woman has a first myocardial infarction is about 75 years, more than two decades later than the onset of menopause. During these years, the risk for breast cancer increases. More women die of coronary artery disease and of complications of osteoporosis than of breast cancer. The optimal time for initiation of hormone replacement therapy and the optimal duration of this therapy are unknown.

Breast Cancer Screening

Screening mammography in older women saves lives and has been shown to be relatively cost-effective. However, whether it is an effective screening test for younger women remains debatable, and one study finds that its cost-effectiveness is low in this context.

Cost-Effectiveness of Early Mammography Was Tested

Salzmann P, Kerlikowske K, Phillips K. Cost-effectiveness of extending screening mammography guidelines to include women 40 to 49 years of age. Ann Intern Med. 1997; 127:955-65.

Experts continue to debate whether women should undergo screening mammography in their 40s. The pooled results of large, randomized trials have shown no reduction in mortality after 7 to 9 years of such screening in women whose screening begins when they are in their 40s, but a statistically significant decrease in mortality has been shown after 10 to 14 years [10, 11]. This study sought to estimate the cost-effectiveness of screening women in their 40s.

The authors created a decision analysis using a hypothetical cohort of women. They based their assumptions about mortality on data from previous trials, and they did sensitivity analyses. For women who begin mammography at 50 years of age, they assumed a mortality reduction of 27% after 5 years. For women who begin mammography at 40 years of age, they assumed a 16% reduction in mortality after 10 years and a 27% reduction after 15 years.

The initial results are shown in Table 5. In a sensitivity analysis, the authors estimated that the cost of screening mammography would need to be $45 for the incremental cost per year of life saved to decrease to $50 000 or less for women 40 to 49 years of age. In contrast, if the true mortality benefit occurs at 12 years, as has been found (instead of at 10 years, as the authors assumed), the incremental cost per year of life saved increases to $200 000 for these women.

Many experts use $50 000 per year of life saved as a rough standard of "acceptable" cost-effectiveness for a screening procedure. It is clear that screening mammography for younger women is much more expensive. However, cost-effectiveness by itself should not dictate whether a test should be performed. Clinicians must understand the reassurance that can be provided by a normal test result and the possible anxiety that can be provoked by a false-positive result. Finally, our society may decide to base this and other decisions about screening on factors other than cost-effectiveness.

Dr. Walsh: University of California at San Francisco/Mount Zion Medical Center, Outpatient Department, 2380 Sutter, Box 1650, San Francisco, CA 94115.

Dr. Nattinger: General Internal Medicine, Medical College of Wisconsin, 8700 West Wisconsin Avenue, Box 135, Milwaukee, WI 53226.

Dr. Roberts (Series Editor): Madrona Medical Group, 3199 Steller Court, Bellingham, WA 98226.